[
https://issues.apache.org/jira/browse/HDFS-9555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15056457#comment-15056457
]
Mingliang Liu commented on HDFS-9555:
-------------------------------------
Thanks for reporting this.
I'm not sure swallowing {{Exception}} instead of {{IOException}} is a good idea
after {{clearCorruptLazyPersistFiles}}, but skipping the sleep logic makes
little sense in the current code.
+1 (non-binding).
> LazyPersistFileScrubber should still sleep if there are errors in the clear
> progress
> ------------------------------------------------------------------------------------
>
> Key: HDFS-9555
> URL: https://issues.apache.org/jira/browse/HDFS-9555
> Project: Hadoop HDFS
> Issue Type: Bug
> Affects Versions: 2.6.0
> Reporter: Phil Yang
> Assignee: Phil Yang
> Attachments: 9555-v1.patch
>
>
> If LazyPersistFileScrubber.clearCorruptLazyPersistFiles throw an exception in
> run(), there will be no sleep logic so it will restart immediately. However
> it may be still fail so there are too many ERROR logs in namenode said
> "Ignoring exception in LazyPersistFileScrubber".
> We need sleep if we catch the exception.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
